Analyzing and Characterizing Heme Binding Peptides

Self-assembling peptides have a variety of uses in biological materials science, from conductive nanowires to applications in biomedicine. This research hopes to facilitate future heme binding ⍺-helical peptide designs by employing machine learning methods to identify patterns in peptide sequences that promote heme binding while maintaining an ⍺-helical structure to mitigate the bias in previous research towards β-sheet forming peptides. Physical and structural properties of the peptides determined from their sequences, including charge distribution, ⍺-helical propensity, heme binding propensity, and hydrophobic interactions motivated the rational design of initial experimental peptide sequences. >200 ixteen-amino-acid-long peptide sequences were synthesised using Solid Phase Peptide Synthesis, and characterized using Fourier Transformation Infrared spectroscopy, Circular Dichroism (CD) spectroscopy and Ultraviolet-visible (UV-Vis) spectroscopy to assess secondary structure and heme binding efficiency. A linear combination analysis routine of the CD and UV/vis data was developed using python to automate and quantify the fidelity of ⍺-helix formation and heme binding. These data will then be used as part of the experimental validation method in our machine learning protocol, consisting of a Long Short Term Memory model that was trained on the Protein Data Bank and experimental data, aimed to generate potential patterns of successful structures for experimentation.